主 题: Marginal Inferential Models: Exact Prior-Free Probabilistic Inference on Interest Parameters
报告人: Prof. Chuanhai Liu (Department of Statistics, Purdue University)
时 间: 2013-12-27 14:00-15:30
地 点: 理科1号楼 1560 （统计中心活动）
Conventional schools of thought on statistical inference are challenged by very-high-dimensional problems and irreproducibility of published research. The Inferential Model (IM) framework, proposed most recently for prior-free probabilistic inference with desirable frequency properties, could perhaps be a promising alternative. The basic idea of IMs is to associate the observed data and unknown quantities/parameters with predictable quantities, called auxiliary variables. IMs produce genuine probabilistic results for scientific inference by predicting unobserved auxiliary variables using predictive random sets, conditioning on observed auxiliary variables.
With a brief introduction to IMs, this talk focuses on inference in the presence of nuisance parameters, which is known as marginal inference. For regular problems, exact and efficient marginalization can be achieved. It is shown that IM approach provides ecient marginal inference in several challenging problems, including a many-normal-means problem, and does not suffer from common marginalization paradoxes. In non-regular problems, a proposed generalized marginalization technique is valid and also paradox-free. Details are given for a benchmark example, namely, the Behrens{Fisher problem.
About the speaker（报告人介绍）：Chuanhai Liu is Professor of Statistics at Purdue University. He received his
MS degree in Probability and Statistics in 1987 from Wuhan University and Ph.D. degree in Statistics in 1994 from Harvard University. Before joining Purdue in 2005, he worked at Bell Laboratories as member of technical staff for ten years. His research interests include modeling, statistical computing, and reasoning with uncertainty (or foundations of statistical inference).
